NEAR-INFRARED and STAR-FORMING PROPERTIES of LOCAL LUMINOUS INFRARED GALAXIES Almudena Alonso-Herrero1, George H

Draft version June 8, 2006 A Preprint typeset using L TEX style emulateapj v. 2/19/04

NEAR-INFRARED AND STAR-FORMING PROPERTIES OF LOCAL LUMINOUS INFRARED GALAXIES Almudena Alonso-Herrero1, George H. Rieke2, Marcia J. Rieke2, Luis Colina1, Pablo G. Perez-Gonz alez 2, and Stuart D. Ryder3 Draft version June 8, 2006

ABSTRACT We use Hubble Space Telescope (HST) NICMOS continuum and Pa observations to study the near-infrared and star-formation properties of a representative sample of 30 local (d 35 75 Mpc) luminous infrared galaxies (LIRGs, infrared [8 1000 m] luminosities of log LIR = 11 11.9 [L]). The data provide spatial resolutions of 25 50 pc and cover the central 3.3 7.1 kpc regions of these galaxies. About half of the LIRGs show compact ( 1 2 kpc) Pa emission with a high surface brightness in the form of nuclear emission, rings, and mini-spirals. The rest of the sample show Pa emission along the disk and the spiral arms extending over scales of 3 7 kpc and larger. About half of the sample contains HII regions with H luminosities signicantly higher than those observed in normal galaxies. There is a linear empirical relationship between the mid-IR 24 m and hydrogen recombination (extinction-corrected Pa) luminosity for these LIRGs, and the HII regions in the central part of M51. This relation holds over more than four decades in luminosity suggesting that the mid-IR emission is a good tracer of the star formation rate (SFR). Analogous to the widely used relation between the SFR and total IR luminosity of Kennicutt (1998), we derive an empirical calibration of the SFR in terms of the monochromatic 24 m luminosity that can be used for luminous, dusty galaxies. Subject headings: galaxies: ISM — ISM: HII regions — galaxies: spiral — infrared radiation — infrared: galaxies — galaxies: interactions — galaxies: star clusters

1. INTRODUCTION bright galaxies show morphologies and properties (see The importance of infrared (IR) bright galaxies has e.g., Zheng et al. 2004; Papovich et al. 2005; Mel- been recognized since their discovery more than 30 years bourne, Koo, & Le Floc’h 2005; Shi et al. 2006) typical ago (Rieke & Low 1972) and the detection of large num- of low-z LIRGs and ULIRGs (among others, Murphy et bers by the IRAS satellite (Soifer et al. 1987). Specif- al. 1996; Surace, Sanders, & Evans 2000, Scoville et al. ically, there has been controversy over the extent to 2000; Alonso-Herrero et al. 2002; Colina et al. 2001; which luminous and ultraluminous IR galaxies (LIRGs, Colina, Arribas, & Clements 2004; Arribas et al. 2004). 11 12 Lagache, Puget, & Dole (2005) have recently reviewed LIR = L[8 1000m] = 10 10 L, and ULIRGs 12 the properties of the high-z IR galaxy population. LIR > 10 L, respectively) in the local universe are powered by intense star formation as opposed to active The local density of LIRGs is two orders of magnitude galactic nuclei (see the review by Sanders & Mirabel higher than that of ULIRGs (e.g., Soifer et al. 1987). 1996). The process of merging with accompanying super- Moreover, Murphy et al. (2001) have argued that in starbursts that produces LIRGs and ULIRGs appears to the local universe interacting galaxies may spend a sig- be an important stage in galaxy evolution, possibly even nicant fraction of their lifetime as LIRGs, whereas the converting spiral galaxies into ellipticals (e.g., Genzel et ULIRG phase may just be short and recurrent during the al. 2001; Colina et al. 2001, and references therein) and merging process. However, the majority of the detailed giving rise to quasars (Sanders et al. 1988). studies of the local universe IR galaxies have been fo- In recent years with the advent of the new generation cused on the objects with the largest IR luminosities (i.e., of IR satellites, there has been a considerable eort in un- ULIRGs). The LIRGs have been generally neglected, ex- derstanding the properties of distant IR-selected galax- cept for detailed studies of a few famous examples (e.g., ies. The majority of these galaxies at z < 1 are in the Gehrz, Sramek, & Weedman 1983; Doyon, Joseph, & LIRG class, and they make a signicant contribution to Wright 1994; Genzel et al. 1995; Satyapal et al. 1999; the star formation rate density at 0.5 < z < 2 (Elbaz Sugai et al. 1999; Lpari et al. 2000, 2004; Alonso- et al. 2002; Perez-Gonzalez et al. 2005; Le Floc’h et al. Herrero et al. 2000, 2001). Thus, understanding low-z 2005). At z 0.7 1 a signicant fraction of the LIRG LIRG samples is critical for interpreting the properties of population appears to be morphologically disturbed spi- IR-selected high-z galaxy populations; in addition, they ral galaxies (Bell et al. 2005; Shi et al. 2006), rather represent the link between ULIRGs and the population than interacting and merging systems. These high-z IR of eld galaxies as a whole. In this paper we present Hubble Space Telescope (HST) 1 Departamento de Astrofsica Molecular e Infrarroja, Instituto NICMOS continuum (1.1, 1.6, and 1.876 m) and Pa de Estructura de la Materia, CSIC, Serrano 121, 28006 Madrid, emission line observations of a volume limited sample Spain; e-mail: [email protected] (30 systems, 34 individual galaxies) of local universe 2 Steward Observatory, The University of Arizona, 933 N. Cherry Avenue, Tucson, AZ 85721 (z < 0.017) LIRGs. The paper is organized as follows. 3 Anglo-Australian Observatory, PO Box 296, Epping, NSW §2 gives details on the sample selection and properties. 1710, Australia In §3 we describe the HST/NICMOS observations, data 2 Local LIRGs

TABLE 1 The sample of local universe LIRGs.

Galaxy Name IRAS Name vhel Dist log LIR Spect. Refs Morphology 1 (km s ) (Mpc) (L) class (1) (2) (3) (4) (5) (6) (7) (8) NGC 23 IRASF 00073+2538 4566 59.6 11.05 H ii 1 paired with NGC 26 MCG +12-02-001 IRASF 00506+7248 4706 64.3 11.44 – – in group? NGC 633 IRASF 013413735 5137 67.9 11.09: H ii 2,3 paired with ESO 297-G012 UGC 1845 IRASF 02208+4744 4679 62.0 11.07 – – isolated NGC 1614 IRASF 043150840 4746 62.6 11.60 H ii 1,2,3 merger UGC 3351 IRASF 05414+5840 4455 60.9 11.22 Sy2 4,5 paired with UGC 3350 NGC 2369 IRASF 071606215 3237 44.0 11.10 – – isolated NGC 2388 IRASF 07256+3355 4134 57.8 11.23: H ii 1 paired with NGC 2389 MCG +02-20-003 IRASF 07329+1149 4873 67.6 11.08: – – in group NGC 3110 IRASF 100150614 5034 73.5 11.31: H ii 2 paired with MCG 01-26-013 NGC 3256 IRASF 102574339 2814 35.4 11.56 H ii 6 merger NGC 3690/IC 694 IRASF 11257+5850 3121 47.7 11.88 Sy2-H ii 7 close interacting pair ESO 320-G030 IRASF 115063851 3232 37.7 11.10 H ii 8 isolated MCG 02-33-098E/W IRASF 125961529 4773 72.5 11.11 H ii (both) 1,3 close interacting pair IC 860 IRASF 13126+2453 3347 59.1 11.17: H ii 5 isolated NGC 5135 IRASF 132292934 4112 52.2 11.17 Sy2 2,3 in group NGC 5653 IRASF 14280+3126 3562 54.9 11.06 H ii 1 isolated NGC 5734 IRASF 144232039 4074 59.3 11.06: NO 1 paired with NGC 5743 IC 4518E/W IRASF 145444255 4715 69.9 11.13 Sy2 (W) 2 close interacting pair Zw 049.057 IRASF 15107+0724 3897 59.1 11.27: H ii 1,5 isolated NGC 5936 IRASF 15276+1309 4004 60.8 11.07 H ii 1 isolated — IRASF 171381017 5197 75.8 11.42 H ii 2,3 isolated? IC 4687/IC 4686 IRASF 180935744 5200/4948 74.1 11.55: H ii (both) 3,9 close interacting pair IC 4734 IRASF 183415732 4680 68.6 11.30 H ii/L 2 in group NGC 6701 IRASF 18425+6036 3965 56.6 11.05 H ii: 1 isolated NGC 7130 IRASF 214533511 4842 66.0 11.35 L/Sy 1,2 peculiar IC 5179 IRASF 221323705 3422 46.7 11.16 H ii 1 isolated NGC 7469 IRASF 23007+0836 4892 65.2 11.59 Sy1 1 paired with IC 5283 NGC 7591 IRASF 23157+0618 4956 65.5 11.05 L 1 isolated NGC 7771 IRASF 23488+1949 4277 57.1 11.34 H ii 1 paired with NGC 7770

Note. — Column (1): Galaxy name. Column (2): IRAS denomination from Sanders et al. (2003). Column (3): Heliocentric velocity from NED. Column (4): Distance taken from Sanders et al. (2003). Column (5): 8 1000 m IR luminosity taken from Sanders et al. (2003), where the sux “:” means large uncertainty (see Sanders et al. 2003 for details). NGC 633 was identied as the IRAS source in Sanders et al. (1995). Interacting pairs for which Surace et al. (2004) obtained IRAS uxes for the individual components: NGC 633 log LIR = 10.56, and for NGC 2388, NGC 7469, and NGC 7771 with values of log LIR similar to those given in Sanders et al. (2003). Column (6): Nuclear activity class from spectroscopic data (“Sy”=Seyfert, “L”=LINER, “H ii”=H ii region-like, “NO”=no classication was possible, “–”=no data available). Column (7): References for the spectroscopic data: 1. Veilleux et al. (1995); 2. Corbett et al. (2003); 3. Kewley et al. (2001); 4. Veron-Cetty & Veron (2001); 5. Baan, Salzer, & LeWinter (1998); 6. Lpari et al. (2000); 7. Garca-Marn et al. (2006). The Sy2 classication is for B1 in NGC 3690; 8. van den Broek et al. (1991); 9. Sekiguchi & Wolstencroft (1992). Column (8): Morphological description. reduction, and continuum and H ii region photometry. systems within this velocity range in the RBGS. In §4 we discuss the near-IR continuum properties of the The sample is presented in Table 1. The IR lu- central regions of local LIRGs. In §5, and §6,we describe minosities (and distances) of the systems are taken the Pa morphology, and the properties of individual from the RBGS (Sanders et al. 2003), and are H ii regions, respectively. In §7 we estimate the extinc- in the range log LIR = 11.05 11.88 [L]. Five tion to the Pa emitting regions. In §8 we estimate the (NGC 1614, NGC 3256, NGC 3690/IC 694, Zw 049.057, extended H emission, and in §9 we discuss the star for- and NGC 7469) of these systems were part of the NIC- mation rates of local LIRGs. Our conclusions are given MOS guaranteed time observations (GTO, see Scoville 1 in §10. Throughout this paper we use H0 = 75 km s et al. 2000; Alonso-Herrero et al. 2000, 2001, 2002), Mpc1. and one (NGC 5653) was part of the NICMOS Snap- shot Survey of Nearby Galaxies (Boker et al. 1999). The 2. THE SAMPLE remaining twenty-four systems are new general observer We have selected a volume limited sample of nearby (GO) observations obtained during HST Cycle 13. LIRGs from the IRAS Revised Bright Galaxy Sample In Table 1 we also list, if available, the nuclear activ- (RBGS: Sanders et al. 2003). The velocity range (vhel = ity class — H ii, Seyfert and LINER — obtained from 2750 5200 km s1) of the galaxies was chosen so that spectroscopic observations in the literature (see Table 1 the Pa emission line (rest = 1.876 m) falls into the notes for references). The galaxies show a variety of mor- HST/NICMOS F190N narrow-band lter. We required phologies and environments, including isolated systems, the logarithm of the total IR luminosity to be 11.05 galaxies in groups, interacting galaxies, and advanced and that the galaxies be at Galactic latitude |b| > 10 deg. mergers. In the case of close interacting galaxies the The sample is composed of 30 systems (34 individual listed IR luminosities correspond to both galaxies in the galaxies) and contains 77% of the complete sample of system. Surace, Sanders, & Mazzarella (2004) presented Alonso-Herrero et al. 3 high resolution IRAS images for close interacting sys- F187N lter. The eld of view (FOV) of the images is tems within the RBGS. Their work contains four sys- approximately 19.500 19.500 which for our sample covers tems in our sample for which the IRAS uxes of the the central 3.37.2 kpc of the galaxies. The observations individual components of the pair are available, as listed were designed such that each galaxy could be observed in the notes to Table 1. There are four further systems within one HST orbit. The typical integration times were for which Surace et al. (2004) could not obtain IRAS 250 s for each of the broad-band lters and 900950 s for uxes for the two individual components: MCG 02-33- each of the narrow-band lters. For each lter the total 098E/W, IC 4518E/W, IC 4686/IC 4687, and Arp 299. integration time was split into three individual exposures The components of Arp 299 have been amply studied in with relative osets of 5 pixels. the mid-IR and their contributions to the total IR lumi- The images were reduced using the HST/NICMOS nosity of the system are well determined (see e.g., Soifer pipeline routines, which involve subtraction of the rst et al. 2001). readout, dark current subtraction on a readout-by- For various reasons, nine galaxies that meet our selec- readout basis, correction for linearity and cosmic ray re- tion criteria have not been imaged in Pa: UGC 2982, jection, and at elding. For a given lter, the individual CGCG 468-002, ESO 264-G057, MCG 03-34-064, images were aligned using common features present in all IC 4280, UGC 8739, ESO 221-G010, NGC 5990, and three images and combined to construct a mosaic. The NGC 7679. We have examined these galaxies carefully nal F110W images, continuum-subtracted Pa images, to see if their properties depart from those in the ob- and F110W F160W color maps are presented in Fig. 1 served sample. They do not. For example, the average for the new observations of galaxies in our sample. logarithm of the luminosity of the observed galaxies is The ux calibration of the images was performed us- log LIR = 11.25 [L], while for the unobserved ones it is ing conversion factors given in the HST/NICMOS hand- log LIR = 11.11 [L]. The average redshift of the former book. We have also examined the near-IR continuum 1 1 group is 4317 km s , and for the latter is 4722 km s . images and the mF110W mF160W color maps (Fig. 1) to The IR spectral shapes are generally similar, except that look for the presence of nuclear point sources. When a the observed sample contains both of the galaxies that nuclear point source was detected we obtained photom- IRAS did not detect at 12 m: IC 860 and Zw 049.057. etry with a 100-diameter aperture (subtracting the un- Within the errors, these galaxies all have approxi- derlying galaxy emission measured in an annulus around mately solar metallicity. Although this conclusion is the point source) and performed an aperture correction expected, we were able to conrm it for all but three (from the NICMOS handbook) to include all the ux members of the sample (MCG +12-02-001, UGC 1845, from the point source. The typical photometry uncer- and IC 860). We did so from measurements of [N ii]/H tainties associated with removing the underlying galaxy and [O iii]/H from the literature, converted to 12 + emission were 0.04 0.06 mag. If a nuclear point source log (O/H) according to the correlations found by Mel- was not detected, we measured the continuum emission bourne & Salzer (2002). For the rst two exceptions, we using a similarly sized aperture but without removing could not nd adequate data, while for IC 860 the emis- the emission from the underlying galaxy. The results for sion line equivalent width is very small and the data in the nuclei are presented in Table 2, and discussed in §4.2. the literature are not consistent. For all the others, the 3.2. ii computed metallicity is within a factor of two of solar. Photometry of H regions Given the approximate approach to the calculations, this The continuum-subtracted Pa line emission images factor is easily within the errors. were produced by subtracting the F187N images from For our selected range of distance, the IRAS sample of the F190N images, both calibrated in Jy. Using the LIRGs should be essentially complete (excluding regions F190N lter we estimate that for galaxies at 5000 . v . of high IR cirrus and those not surveyed). Our observed 5200 km s1 up to 20% (for the typical widths of hydro- sample is 77% complete in terms of the IRAS sample, and gen recombination lines of LIRGs not containing a type the missing galaxies are very similar to those observed. 1 AGN, see Goldader et al. 1997a) of the Pa ux could We conclude that our results will be representative of be lost due to the observed wavelength of the emission local LIRGs in general. line (see discussion in Alonso-Herrero et al. 2002). We have used SExtractor (Bertin & Arnouts 1996) 3. HST/NICMOS OBSERVATIONS to analyze the properties of the individual H ii regions In this section we only describe the new HST/NICMOS detected in the Pa images of our sample of galaxies. observations obtained in Cycle 13. We refer the reader We have set the lower limit for the size of an H ii region to Scoville et al. (2000) and Alonso-Herrero et al. (2000, to 9 contiguous pixels. For the distances of the galaxies 2001, 2002) for details on the GTO observations, data in our sample, this corresponds to minimum linear sizes reduction, and ux calibration of the data obtained prior (equivalent diameters) of between 44 and 96 pc. The de- to Cycle 13 (see §2). tection threshold criterion for a pixel to be included as ii 3.1. part of an H region is to be above the background plus Data reduction twice the rms noise of the background. SExtractor The observations of the LIRGS were taken with the constructs a background image by a bi-cubic interpola- NIC2 camera (plate scale of 0.07600pixel1) using the tion over areas of the background with a size specied by F110W and F160W broad-band lters, and the F187N the user. These areas ought to be larger than the typical and F190N narrow-band lters. At the distances of the size of an H ii region, but not so large as to smooth over LIRG sample, the F190N lter contains the Pa emission local variations of the background. The main source of line and the adjacent continuum at 1.90 m. For contin- uncertainty in measuring the Pa uxes of the H ii re- uum subtraction, we used the images taken through the gions is the background removal. After modelling the 4 Local LIRGs

Fig. 1.— HST/NICMOS 1.1 m continuum emission images (F110W lter, left panels), continuum-subtracted Pa line emission images (F190N-F187N lters, middle panels), and mF110W mF160W color maps (right panels, the scale on the right hand side of the images is the mF110W mF160W color in magnitudes) for the galaxies in our sample observed during Cycle 13. The arrow indicates north, and east is in the counterclockwise direction. Alonso-Herrero et al. 5

Fig. 1.— Continued. 6 Local LIRGs

Fig. 1.— Continued. Alonso-Herrero et al. 7

Fig. 1.— Continued. 8 Local LIRGs

0.5cm

Fig. 1.— Continued. Alonso-Herrero et al. 9

Fig. 1.— Continued. 10 Local LIRGs

Fig. 1.— Continued.

background and H ii regions, we output their positions, incomplete samples or have been focused on the most lu- and isophotal uxes, errors, and areas after subtracting minous objects of this class (e.g., Wu et al. 1998; Arribas the global background from the observed Pa ux. et al. 2004). The properties of the H ii regions detected in the The present study extends the previous high-resolution LIRGs are summarized in Table 3. We list the H lumi- near-IR morphological studies of LIRGs and ULIRGs nosity (the Pa/H ratio varies by about 10% for various (see Scoville et al. 2001, and Bushouse et al. 2002) to conditions and under Case B; we take a ratio of 8.6 from a representative local sample where most galaxies are at Hummer & Storey 1987) for the median H ii region, the the lower luminosity end of the LIRG range. In our sam- brightest H ii region, and the average H luminosity of ple, approximately 50% of the systems are pairs, but only the three brightest H ii regions (rst-ranked H ii regions), a small fraction are closely interacting/merging galaxies as well as the ratio of the sum of the luminosities of the (see Table 1). three brightest H ii regions to the observed H luminos- The most common continuum features in our sample ity in H ii regions, and the observed H luminosity in are luminous star clusters and large-scale spiral arms. H ii regions. These properties are discussed in §6. Two-armed spiral arm structures are also observed extending down to the center (inner kpc scale, the so-called 4. NEAR-IR CONTINUUM PROPERTIES OF THE nuclear dusty spirals, see Martini et al. 2003). These CENTRAL REGIONS OF LIRGS spiral structures (both large and small scale) are also 4.1. Morphology present in at least one galaxy member of close interacting Morphological studies of complete samples of local systems or in mergers (NGC 1614, IC 694, and IC 4687), ULIRGs have found that the great majority appear to and pairs of galaxies. In closely interacting systems a be merging and/or disturbed systems (e.g., Murphy et more perturbed central morphology would be expected al. 1996; Clements et al. 1996). At lower IR luminosities (and observed, see Scoville et at 2000 and Bushouse et (the LIRG category) morphological studies have shown al. 2002), although this depends on a number of factors, similarities with the ULIRG class although the fraction including relative sizes, and amount of gas present in the of strongly interacting/disturbed systems appears to be involved galaxies, and stage of the interaction. smaller, approximately 5060% (Wu et al. 1998). How- In highly inclined galaxies in our sample dust lanes ever, these LIRG studies have often been conducted for crossing the disks are common and present the reddest Alonso-Herrero et al. 11 mF110W mF160W colors (see Fig. 1). If these dust lanes are hiding an old stellar population (see Table 2), the typical extinctions would be in excess of AV = 34 mag, using the Rieke & Lebofsky (1985) extinction law. The most “extreme” morphological features commonly observed in more IR luminous systems (see Scoville et al. 2000 and Bushouse et al. 2002), such as double/multiple nuclei systems with a large number of bright star clusters near their centers and star-forming regions in the interface of interacting galaxies, are not very common in our sample of LIRGs. The most representative examples in this class are NGC 3690 (see Alonso-Herrero et al. 2000) and NGC 3256 (see Alonso-Herrero et al. 2002 and Lpari et al. 2004). These two systems are among the most IR luminous objects in our sample. In terms of the central dust distribution, a few galaxies are dominated by nuclear extinction: IC 860, MCG 02- 33-098E/W, IC 4518W, and IRAS 171381017, whereas in other galaxies the regions of the largest extinction do not coincide with the nucleus of the galaxy (e.g., NGC 3110, NGC 5734). In the cases of galaxies with nuclear rings (see §5.3) of star formation, the H ii regions are interleaved with the dust features or the color maps trace a ring of extinction just outside the ring of H ii re- Fig. 2.— Left panels: H luminosity of the median H ii region as a function of the IR luminosity (bottom) and total observed (not gions (e.g., NGC 23, and NGC 1614, see Alonso-Herrero corrected for extinction) H luminosity of the galaxy (top) in H ii et al. 2001). regions in the central regions. The horizontal dashed line shows the H luminosity of 30 Doradus in the LMC, the prototypical 4.2. Nuclei of Galaxies giant H ii region. The error bar indicates the typical standard deviation from the mean value of the distribution of H ii region H Scoville et al. (2000) and Bushouse et al. (2002) found luminosities of a given galaxy Right panels: Average H luminosity near-IR point-like nuclear sources in approximately 30 of the three brightest (rst-ranked) H ii regions as a function of 50% of the galaxies in their samples of IR bright galax- the IR luminosity (bottom) and total observed (not corrected for ies (i.e., some LIRGs but mostly ULIRGs). Scoville et extinction) H luminosity of the galaxy (top). al. (2000) established that in about one-third of their sample these nuclear point sources dominate the near-IR emission at 2.2 m. These bright nuclear point sources Herrero et al. 2000), or in the case of the compact galaxy are more prevalent in LIRGs and ULIRGs with Seyfert IC 4686. or QSO activity. This is not surprising, as bright near- The majority of the nuclei display redder colors (see IR nuclear point sources are common in nearby Seyfert 1 Table 2) than those typical of an old stellar population, galaxies, and are present in about 50% of nearby Seyfert implying extinctions of a few magnitudes (or higher if 2 galaxies (see Quillen et al. 2001). However, galaxies the stellar population is young, see §7.2) for a simple with no evidence of AGN activity can also show nuclear dust screen model. Scoville et al. (2000) found values point sources, believed to be nuclear star clusters. The of the nuclear mF110W mF160W colors of between 1 typical luminosities of the nuclear star clusters in late and 2 for both LIRGs and ULIRGs in their sample, thus 6 7 type spiral galaxies are in the range 10 10 L (Boker similar to the colors observed in our sample. Scoville et et al. 2004 and references therein), similar to those of the al. (2000) also found that the mF160W mF222M colors so-called super star clusters, and more luminous galaxies tend to be redder for the ULIRGs than the less luminous tend to harbor more luminous nuclear clusters. LIRGs in their sample. We do not see any clear evidence We have found that approximately 70% of the galaxies for this behavior, although our sample, unlike that of in our sample (including the galaxies observed prior to Scoville et al. (2000), does not contain examples of the Cycle 13) show nuclear point sources, with observed (not most extreme cases of ULIRGs hosting QSOs, known to corrected for extinction) absolute H-band magnitudes present very red near-IR colors. ranging from MF160W = 22.9 (for the bright Seyfert 5. 1 nucleus in NGC 7469) to MF160W = 18.8, with an MORPHOLOGY OF THE PA LINE EMISSION OF THE CENTRAL REGIONS OF LIRGS average value of MF160W = 20.3 (see Table 2). The observed absolute H-band magnitudes for the resolved The HST/NICMOS Pa observations presented in this central regions (for sizes 170 370 pc, depending on the study aord us the best opportunity to study the H ii re- distance) are similar to those of the nuclear point sources. gion emission with spatial resolutions of 25 50 pc in a In general, the observed range of F160W absolute mag- sample of local LIRGs. The advantage of the Pa obser- nitudes (Table 2) does not allow us to discriminate be- vations when compared to optical H observations is that tween AGN and star clusters. We also nd that these the extinction is reduced by a factor of approximately 5 point sources do not dominate the near-IR emission ex- (APa ' 0.18AH, using the Rieke & Lebofsky 1985 except in some of the most IR luminous galaxies (e.g., at tinction law). In very dusty systems, such as those under 2.2 m the Seyfert 1 nucleus of NGC 7469 see Scoville consideration here, high spatial resolution mid-IR imag- et al. 2000, and component B1 in Arp 299 see Alonso- ing has shown the location (nuclear regions) and extent 12 Local LIRGs

TABLE 2 Observed photometry of the nuclei of LIRGs.

Galaxy mF160W mF110W mF160W fF187N/fF160W MF160W (1) (2) (3) (4) (5) Nuclear point sources NGC 23 13.26 1.07 0.99 20.6 NGC 633 14.78 1.14 1.03 19.4 MCG +12-02-001 13.61 1.37 1.15 20.4 UGC 1845 12.41 1.48 1.21 21.6 NGC 2388 13.07 1.26 1.13 20.7 MCG 02-33-098W 14.26 1.22 1.19 20.0 IC 860 14.43 1.54 1.25 19.6 NGC 5135 14.46 1.45 1.54 19.1 NGC 5734 13.51 1.11 0.98 20.4 IC 4518W 14.40 1.95 1.74 19.8 IC 4686 13.29 0.92 1.01 21.1 NGC 6701 13.68 1.10 1.02 20.1 IC 5179 14.00 1.32 1.17 19.4 NGC 7771 14.99 1.67 1.32 18.8 Nuclear regions UGC 3351 13.97 1.93 1.30 20.0 NGC 2369 14.23 1.59 1.30 19.0 MCG +02-20-003 13.94 1.54 1.29 20.2 NGC 3110 14.19 1.19 1.11 20.1 ESO 320-G030 13.62 1.28 1.11 19.3 MCG 02-33-098E 14.64 1.30 1.19 19.7 IC 4518E 14.66 1.44 1.21 19.6 IRAS 171381017 14.74 1.87 1.38 19.7 IC 4687 14.24 1.33 1.14 20.1 IC 4734 13.27 1.43 1.19 20.9 NGC 7130 13.56 0.95 1.11 20.5 NGC 7591 13.49 1.43 1.21 20.6 Note. — Column (1): Galaxy name. Column (2): Observed NICMOS F160W magnitude. Column (3): Observed mF110W mF160W color. The typical color of an old stellar population is mF110W mF160W = 1.03, based on the integrated spectrum of an elliptical galaxy using the lter transmissions and the quantum eciency curves for the NICMOS lters. Column (4): 1.87 m to 1.60 m continuum ux ratio. For the same elliptical galaxy as in Column (3), the 1.87 m to 1.60 m ux density ratio is fF187N/fF160W ' 0.85. Column (5): Absolute NICMOS F160W magnitude. Not obvious nucleus, the aperture was centered at the brightest F187N source. In the case of IRAS 171381017 this corresponds to the northern nucleus (see Zhou, Wynn-Williams, & Sanders 1993). In addition, near-IR nuclear point sources are present in NGC 7469 (Seyfert 1 nucleus: MF160W = 22.9, Scoville et al. 2000), NGC 1614 (MF160W = 22.3), NGC 3690 (B1: MF160W = 19.5 and B2: MF160W = 20.2), NGC 5653 (MF160W = 20.8), and NGC 3256 (MF160W = 20.2).

(approximately on scales of 1 kpc or less) of the current and Martini et al. 2003). The typical observed H sur- dusty star formation (Soifer et al. 2001). It is of interest face brightnesses are similar to the galaxies with compact to determine if the Pa emission is able to penetrate the Pa emission. high column densities traced by the mid-IR emission. We have grouped the main morphological Pa features 5.3. Nuclear star-forming rings present in the LIRG sample (see middle panels of Fig. 1) The nuclear rings in our sample of LIRGs (Table 3) in ve categories. We note that some galaxies show mor- have diameters ranging from 0.7 kpc to 2 kpc, show re- phologies that place them in more than one category (see solved H ii regions within them, and dominate the central Table 3). Pa emission in these galaxies. Fainter H ii regions are 5.1. also present in the spiral arms in these systems. The typ- Compact (. 1 kpc) nuclear Pa emission ical observed (not corrected for extinction) H surface The Pa morphology appears dominated by the nu- brightnesses in the H ii regions are: 2 4 1041 erg s1 clear emission (see Table 3 for the galaxies in this class) kpc2, although the ring of star formation in NGC 1614 with sizes of less than approximately 1 kpc, although has an H surface brightness of ' 601041 erg s1 kpc2. fainter H ii regions are also present outside of the nuclear All these nuclear rings of star formation except the one regions. In some cases this compact emission appears to in ESO 320-G030 are located in interacting galaxies. be related to the presence of an AGN (see Table 1, and also §9.2.2). The average observed H surface bright- 5.4. Large-scale (several kpc) Pa emission with H ii nesses (derived from Pa) are among the highest in our regions located in the spiral arms sample of galaxies: 5 10 1041 erg s 1 kpc 2. In this category we nd LIRGs with H ii regions located 5.2. over several kpc along the spiral arms of the galaxy (see Nuclear mini-spiral (inner 1 2 kpc) Pa emission Table 3). Although our images only cover the central In these cases we observe bright H ii regions as well as ' 3.3 7.2 kpc, several of these galaxies are known to bright star clusters along a nuclear mini-spiral structure show H emission over larger scales (see §8, and Lehn- (see Table 3) also detected in continuum light (see §4.1, ert & Heckman 1995; Dopita et al. 2002; Hattori et al. Alonso-Herrero et al. 13

TABLE 3 Statistics of the H ii regions in the central regions of LIRGs.

Galaxy Pa log L(H)median log L(H)br log L(H)3 L(H)3/L(H)HII L(H)HII morph. erg s1 erg s1 erg s1 1040erg s1 (1) (2) (3) (4) (5) (6) (7) NGC 23 R 39.53 40.94 40.90 0.24 99.6 MCG +12-02-001 M 39.50 41.96 41.60 0.61 199.4 NGC 633 R 39.16 41.17 40.88 0.63 36.5 UGC 1845 M 39.32 41.23 41.03 0.39 81.7 UGC 3351 E 39.20 40.71 40.60 0.35 34.1 NGC 2369 E 38.96 40.87 40.80 0.39 48.6 NGC 2388 E 39.33 41.66 41.26 0.67 81.4 MCG +02-20-003 C 39.25 41.31 40.96 0.69 39.4 NGC 3110 S 39.60 40.90 40.82 0.30 64.8 ESO 320-G030 R 39.47 40.31 40.28 0.13 43.6 MCG 02-33-098 C 41.76 0.95 101.6 IC 860 C 40.13 0.90 1.9 NGC 5135 M 39.49 41.05 41.00 0.36 83.1 NGC 5734 S 39.19 40.52 40.28 0.19 29.6 IC 4518W C 39.52 41.28 41.16 0.69 62.4 IC 4518E E 40.21 40.16 0.40 10.8 NGC 5936 M, S 39.23 41.24 41.05 0.68 50.3 IRAS 171381017 M 39.63 41.24 41.10 0.38 99.9 IC 4687 S 39.76 41.42 41.22 0.24 211.0 IC 4686 C 41.63 0.91 82.1 IC 4734 C, S 39.20 41.02 40.99 0.54 54.3 NGC 6701 M 39.64 40.91 40.78 0.40 45.1 NGC 7130 S 39.34 41.69 41.30 0.74 80.4 IC 5179 S 39.24 41.18 40.81 0.39 49.0 NGC 7591 C, S 39.24 41.06 40.80 0.85 22.1 NGC 7771 R, E, S 39.62 41.05 40.90 0.31 76.0 NGC 1614 R 39.88 40.99 40.98 0.08 380.2 NGC 3256 S 39.47 41.49 41.20 0.40 120.2 NGC 3690 S 39.83 41.54 41.44 0.44 186.2 IC 694 S 39.67 41.82 41.40 0.66 114.8 NGC 5653 S 39.62 41.15 40.87 0.23 97.7 Zw 049.057 E 41.32 ' 1 20.9 Note. — Column (1): Galaxy Name. Column (2): Pa morphology described in §5 . C=compact emission. M=nuclear mini-spiral. R=ring of star formation. S=H ii regions in spiral arms. E=edge-on galaxy. Column (3): H luminosity of the median H ii region for galaxies with more than twenty H ii regions detected. The typical standard deviations from the mean value of the distribution of H luminosities of the H ii regions detected is 0.4 0.6 dex for a given galaxy. Column (4): H luminosity of the brightest H ii region. Column (5): Average H luminosity of the brightest three (rst-ranked) H ii regions. The typical uncertainties for the sextractor uxes for the brigthest and rst-ranked H ii regions are 1 2%. Column (6): Ratio of the sum of the H luminosities of the brightest three H ii regions to the H luminosity in H ii regions. Column (7): Observed H luminosity in H ii regions in the central regions. The rst part of the table are the galaxies observed in Cycle 13, whereas the second part are the properties of the galaxies observed prior to Cycle 13 and presented in Alonso-Herrero et al. (2002), recomputed for the distances used in this paper.

2004). The observed H surface brightnesses in H ii re- over scales of a few kpc (3.3 to 7.2 kpc, at least, see §8). gions are about one order of magnitude fainter than those This category is similar to that discussed in §5.4 but for of galaxies with compact Pa emission. galaxies viewed edge-on. These galaxies tend to show the In some cases the nuclear compact Pa emission reddest mF110W mF160W central colors over the extent is bright and can make a signicant contribution to of the Pa emission (see right panels of Fig. 1). the total H ii region emission, for instance in IC 694 Summarizing, over half the galaxies in our sample most and NGC 3690 (see Alonso-Herrero et al. 2000) and of the Pa emission is more compact than the continuum NGC 3256 (see Alonso-Herrero et al. 2002). In this near-IR emission, and is located in the inner 1 2 kpc class we also nd examples of galaxies without bright of the galaxies, as also shown by mid-IR observations nuclear emission and large numbers of bright H ii regions (Soifer et al. 2001) of a few bright LIRGs. In a few in the spiral arms (IC 4687), and even galaxies whose cases the nuclear emission even dominates the total Pa Pa emission is dominated by one or a few bright extra- emission (see further discussion in §8). However, in the nuclear H ii regions: NGC 5653 (see Boker et al. 1999 other half of our sample the Pa emission is extended and Alonso-Herrero et al. 2002) and NGC 5734. on scales larger than a few kpc (as discussed in §5.4 and §5.5). 5.5. Large-scale Pa emission in highly inclined dusty systems

In these cases (see Table 3) the Pa emission traces the 6. least obscured H ii regions along the disk of the galaxies PROPERTIES OF THE H II REGIONS IN LIRGS 14 Local LIRGs

TABLE 4 Pa and H fluxes, and extinctions.

Galaxy Aperture f(H) f(Pa) AV AV (Veilleux) erg cm2 s1 erg cm2 s1 mag mag (1) (2) (3) (4) (5) (6) NGC 23 200 700 (E-W) 2.5 1013 7.64 1014 1.7 0.4 2.3 NGC 3110 200 600 (E-W) 7.2 1014 4.64 1014 3.1 0.4 2.8 MCG 02-33-098E 200 600 (N-S) 3.0 1014 7.22 1014 5.5 0.4 4.5 MCG 02-33-098W 200 600 (N-S) 4.4 1014 1.05 1013 5.5 0.4 2.4 NGC 5734 200 700 (E-W) 2.8 1014 3.48 1014 4.3 0.4 3.6 NGC 5936 200 700 (E-W) 8.9 1014 1.11 1013 4.3 0.4 4.7 NGC 6701 1.500 700 (E-W) 1.0 1013 6.32 1014 3.1 0.4 3.9 NGC 7130 1.500 600 (E-W) 2.1 1013 1.10 1013 2.7 0.4 3.3 IC 5179 200 900 (E-W) 4.4 1014 8.69 1014 5.1 0.4 3.7 NGC 7591 1.500 600 (E-W) 2.4 1014 4.42 1014 5.0 0.4 4.3 NGC 7771 1.500 700 (E-W) 5.9 1014 8.28 1014 4.5 0.4 6.3

Note. — Column (1): Galaxy Name. Column (2): Extraction aperture of the optical spectroscopy. Column (3): Observed H ux from Veilleux et al. (1995). “” next to the H uxes indicates that the optical spectra were obtained under non-photometric conditions (Kim et al. 1995). Column (4): Observed Pa ux for the same aperture. Column (5): Spectroscopic extinction derived from the H and Pa uxes. The errors are computed for a 15% uncertainty in the Pa emission line ux measurements. Column (6): Spectroscopic extinction derived from H and H by Veilleux et al. (1995) for the same extraction apertures.

Alonso-Herrero et al. (2002) showed for a sample of There is a good correlation between the H luminos- seven4 LIRGs that these galaxies contained a signicant ity of the median H ii regions and the IR and total ob- number of exceptionally bright H ii regions (based on the served H luminosity of the system (left panels of Fig. 2). median values of the H ii region luminosity distribution) These correlations appear to be linear. We also show with H luminosities comparable to that of the giant H ii in these gures the typical standard deviation from the region 30 Doradus (log L(H) = 39.70 [erg s1], Kenni- mean value of the distribution of H luminosities of the cutt, Edgar, & Hodge 1989). This small sample was, detected H ii regions for a given galaxy, an indication however, heavily biased towards the most IR luminous of the width of the distribution. Similarly, there is a objects in the LIRG class. Table 3 summarizes the statis- good relation between the luminosity of the rst-ranked tical properties of the H ii regions detected in our sample regions and the IR and the total H luminosity in H ii of LIRGs. regions (right panels of Fig. 2). The uncertainties for Conrming the results of Alonso-Herrero et al. (2002), the sextractor H uxes of the rst-ranked H ii re- the LIRGs in our sample with the highest IR and ob- gions (1 2%) plus those of the distance determinations served H luminosities have median H ii regions with ( 10%) of the galaxies do not account for the dispersion H luminosities comparable or brighter than the giant in H luminosity of the rst-ranked (and also median) H ii region 30 Dor (Table 3, and left panels of Fig. 2). H ii regions for a given IR luminosity. This dispersion is Approximately 50% of our volume-limited sample of probably due to resolution eects, as the distances of our 1 LIRGs have log L(H)median > 39.5 [erg s ], signi- galaxies span a factor of two. cantly higher (about one order of magnitude) than the median H ii regions of normal galaxies in the Virgo Clus- 7. EXTINCTION CORRECTION OF PA MEASUREMENTS ter observed with similar spatial resolutions (see Alonso- We now prepare to compare the star formation rates Herrero & Knapen 2001). (SFR) derived from the IR luminosity and from the num- The rst-ranked (the brightest three) H ii regions ber of ionizing photons (N ) (see §9). To do so, we must are usually coincident with the nuclei of the galax- Ly correct the observed hydrogen recombination line emis- ies, although with notable exceptions, e.g., NGC 5653, sion for extinction (see discussion in Kennicutt 1998). NGC 5734, and IC 4687, as well as those galaxies with Since the mid-IR emission of (some) LIRGs is predomi- nuclear rings of star formation (see §5.3). The average nantly produced in nuclear starbursts with sizes of less H luminosities of the rst-ranked H ii regions are be- than approximately 12 kpc (see e.g., Wynn-Williams & tween 1 and 2 orders of magnitude brighter than the Becklin 1993; Miles et al. 1996; Keto et al. 1997; Soifer median H ii regions, and are well above that of 30 Dor et al. 2001), there must be large concentrations of dust (Table 3). The contribution of the rst-ranked H ii re- in their nuclear regions. gions to the total observed H luminosity in H ii regions The largest uncertainty in determining the extinction in the central regions is given in Table 3. Not surpris- by far comes from the unknown distribution of dust ingly, this contribution is highest for the galaxies with within the source. The color maps (Fig. 1) of most compact Pa emission where these reions also tend to LIRGs in our sample show complicated dust features in coincide with or are near the nucleus of the galaxy. their centers. In the cases of the deeply embedded star- 4 Two galaxies in their sample, VV 114 and NGC 6240, are at forming regions, common in LIRGs and ULIRGs, a sim- v > 5200 km s1, and thus are not included in our sample. ple foreground dust screen model may not provide a good t to the star-forming properties (e.g., see Genzel et al. Alonso-Herrero et al. 15

TABLE 5 Pa and Br fluxes, and extinctions.

Galaxy Aperture f(Br) f(Pa) AV erg cm2 s1 erg cm2 s1 mag (1) (2) (3) (4) (5) NGC 2388 300 900 (E-W) 26.8 1015 2.02 1013 13 5 NGC 3110 300 900 (E-W) 6.0 1015 0.68 1013 2 5 NGC 3256-nucleus 3.500 3.500 54.0 1015 5.37 1013 6 3 NGC 3256-500 S 3.500 3.500 15.0 1015 1.15 1013 15 5 NGC 3256-500 E 3.500 3.500 17.0 1015 NGC 5135 300 1200 (E-W) 16.5 1015 1.95 1013 0.5 5 IRAS 171381017 300 900 (N-S) 22.4 1015 1.62 1013 14 5 NGC 7130 1.500 4.500 (E-W) 10.2 1015 1.00 1013 7 5

Note. — Column (1): Galaxy Name. The two other galaxies in common with Goldader et al. (1997a) are NGC 1614 analyzed in Alonso-Herrero et al. (2001), and NGC 7469 for which the spectroscopic data includes the Seyfert 1 nucleus. Column (2): Extraction aperture. Column (3): Observed Br ux from Goldader et al. (1997a) for all galaxies except for NGC 3256 for which the data are from Doyon et al. (1994). For NGC 3256 the 500 to the south location coincides with the secondary nucleus of the galaxy, and the 500 to the east location, an H ii region, is not fully covered by our Pa image. Column (4): Observed Pa ux for the same aperture. Column (5): Derived spectroscopic extinction. The errors are computed for a 15% uncertainty in the Pa emission line ux measurements.

1995; Genzel et al. 1998; Satyapal et al. 1999), and a or slightly smaller than the AV obtained from H and model in which stars and dust are mixed will be more Pa. The extinctions derived using the longest wave- realistic (Witt & Gordon 2000). Moreover, the simple length emission lines (Table 5) tend to be higher for the model of a foreground dust screen only provides a lower few galaxies in common with H and Pa measurements. limit to the true obscuration, and in cases of high optical The average extinctions obtained from H and Pa over depths both models will only measure the obscuration to regions with typical sizes of 0.5 kpc 1.5 kpc are not too the material less aected by extinction (Goldader et al. dierent from the AV (derived using the same emission 1997a). Because of this, in very dusty systems there is a lines) of individual H ii regions in spiral galaxies and star- tendency for the derived extinction to increase as longer burst galaxies (see e.g., Quillen & Yukita 2001; Maoz et wavelengths are used (see e.g., McLeod et al. 1993; Gen- al. 2001; Scoville et al. 2001; Calzetti et al. 2005). zel et al. 1998). Finally there is the issue of whether the extinction to the stars is the same as the extinction to the 7.2. Extinction to the stars gas. Calzetti, Kinney, & Storchi-Bergmann (1994) found that in general the extinction to the gas (ionized by the We can alternatively use the observed near-IR colors young stars) is twice that of the continuum (mostly pro- to estimate the extinction to the young stars responsi- duced by old stars), implying that the youngest ionizing ble for ionizing the H ii regions. The near-IR colors in stars are located in dustier regions than the cold stellar general present a smaller dependence on the age of the population is. The lack of correspondence between gas stellar population than the optical colors. However, there emission and continuum emission generally observed in is a weak dependence, in particular for the youngest stel- LIRGs (Fig. 1) suggests that this may also be the case lar populations responsible for ionizing the H ii regions. for them. Alonso-Herrero et al. (2002) found, for a small sample of LIRGs, that most young star-forming regions (with Pa 7.1. emission) in LIRGs have ages of ' 1 9 Myr (depending Extinction to the gas on the assumed type of star formation). Kim et al. (1995) and Veilleux et al. (1995), and Using the typical near-IR colors of a young ionizing Goldader et al. (1997a,b) presented the most compre- stellar population predicted by the Rieke et al. (1993) hensive optical and near-IR long-slit spectroscopic sur- models, and the observed F110W-F160W color maps veys, respectively, of the central regions of LIRGs. We (right panels of Fig. 1), we have derived 2-D extinction have a number of galaxies in common with these stud- (refered to as photometric AV ) maps. The uncertainties ies for which we can estimate the extinction to the gas. of the extinction maps are given by the range of pos- We have simulated the sizes and orientations of the slits sible ages of the Pa emitting regions. The photomet- used by Kim et al. (1995) and Goldader et al. (1997a) to ric AV maps are only applicable to continuum regions measure the Pa uxes and compared them with their with Pa ux emission. For each galaxy, the extinction- H and Br measurements, respectively. We have then corrected Pa maps were constructed using the AV maps estimated the extinction to the gas using the Rieke & and the observed Pa images. To estimate the observed Lebofsky (1985) extinction law and a foreground dust and extinction-corrected Pa uxes, we only added those screen model. In addition, for NGC 3256 we have used pixels above the threshold used for estimating the uxes the Br uxes of Doyon et al. (1994). of H ii regions in §6. An estimate of the background is The results are presented in Tables 4 and 5. Table 4 obtained from those pixels below the threshold value. We additionally lists the AV derived from H and H by note that in computing the observed Pa uxes in this Veilleux et al. (1995), which are generally similar to fashion we are including more pixels than when measur- 16 Local LIRGs

TABLE 6 TABLE 7 Photometric extinctions, extinction-corrected Pa Large-scale emission. luminosities, and SFRs. fNIC2 FOV Galaxy Diameter ftot Galaxy Phot AV log L(Pa)corr log SFR IR/NLy H H 24 m 1 mag erg s (1) (2) (3) (1) (2) (3) (4) NGC 23 55 0.70 0.83 +1.1 NGC 23 2.70.8 41.32 0.14 (0.01) UGC 3351 0.73 +1.0 NGC 3110 76 0.50 0.63 MCG +12-02-001 4.50.8 41.67 0.17 +1.0 NGC 6701 0.73 NGC 633 2.60.8 40.91 0.10 +1.1 IC 5179 0.61 UGC 1845 4.60.8 41.33 0.14 NGC 7771 93 0.44 0.55 +1.0 UGC 3351 5.90.8 41.25 0.38 (0.23) +1.1 NGC 2369 4.80.9 41.16 0.34 Note. — Column (1): Galaxy. Column (2): Diameter in +1.1 arcsec used to measure the total H emission. Column (3): NGC 2388 3.50.8 41.29 0.35 +1.0 Ratio of the observed H emission over the NICMOS NIC2 MCG+02-20-003 3.30.8 41.16 0.33 00 00 +1.1 FOV ( 20 20 ) and the total observed H emission using NGC 3110 3.30.8 41.26 0.45 (0.15) the images of Hattori et al. (2004). Column (4): Ratio of the +1.1 ESO 320-G030 3.30.8 40.97 0.53 Spitzer/MIPS 24 m emission over the NICMOS FOV and the +1.1 entire galaxy. MCG02-33-098 3.30.8 41.32 0.20 +0.8 IC 860 3.20.7 39.86 1.72 +1.1 NGC 5135 3.30.9 41.26 0.32 +1.2 NGC 5734 3.10.9 41.05 0.39 +0.8 metric extinction measured over larger regions will tend IC 4518W 4.10.6 41.27 0.17 +1.2 to be smaller than the spectroscopic values. We also nd IC 4518E 4.60.9 40.72 0.12 +1.3 that the highest values of the average photometric extinc- NGC 5936 3.70.9 41.20 0.27 +0.7 tion over the Pa emitting regions tend occur in those IRAS 171381017 5.10.8 41.48 0.34 +1.2 galaxies with the most compact (' 1 kpc) Pa emitting IC 4687 3.10.9 41.63 0.19 +1.0 regions (e.g., MCG +12-02-001) or in the edge-on sys- IC 4686 2.20.7 41.21 0.15 +1.2 tems (e.g., NGC 2369, UGC 3351). IC 4734 3.90.9 41.19 0.52 +1.1 NGC 6701 3.00.9 41.04 0.42 (0.27) 8. LARGE-SCALE EMISSION +1.2 NGC 7130 2.90.9 41.30 0.45 +1.2 As we saw in §5 about half of the galaxies in our sample IC 5179 3.30.8 41.22 0.35 (0.13) +1.2 show extended Pa emission covering the entire FOV of NGC 7591 4.10.9 41.09 0.36 +1.2 the NICMOS observations. Before we compare the Pa NGC 7771 4.10.9 41.36 0.39 (0.04) (or H) and IR luminosities, we need to assess the importance of the H ii region and diuse emission at large Note. — Column (1): Galaxy name. Column (2): Average photo- galactocentric distances, especially for the nearest exam- metric extinction AV over the Pa emitting region derived from the observed mF110W mF160W colors (see §7.2). The uncertainties of ples in our sample. Hattori et al. (2004) for their sam- phot AV are for the possible range of ages of the Pa emitting regions ple of 60 m ux-limited LIRGs, which includes some (4 Myr and 9 Myr) using the Rieke et al. (1993) models for a Gaussian of the galaxies in our sample, have found that the star burst with FWHM=5 Myr (see Alonso-Herrero et al. 2002 for details). We also list the Pa luminosities over the NICMOS FOV corrected formation activity is dominated by emission extending for extinction. and uncertainties. Column (3) Pa luminosity cor- over several kpc. We have obtained the H images from rected for extinction. Column (4): Log of the ratio between the IR Hattori et al. (2004) for the galaxies in common with and corrected for extinction NLy SFRs calculated using the prescrip- our work, and measured the H uxes within the NIC- tions of Kennicutt (1998). In brackets we give the values corrected for extended emission (§8, and Table 7). MOS FOV and the total extent of the galaxy. As can be seen from Table 7, the contribution from the large-scale H emission can be as high as 50% in some galaxies. We have eight further galaxies in common with the sam- ing H ii region uxes with SExtractor (§3.3), as we are ple of Dopita et al. (2002). From inspection of their not imposing a criterion for dening the size of the emit- H images, which comprise the entire systems, we infer ting regions. The typical uncertainties for the summed that the NICMOS Pa images include most of the emis- Pa uxes are ' 30%, depending on the threshold value sion of NGC 633, MCG02-33-098E/W, IC 4518E/W, used for adding up the emission. The average photo- NGC 5734, and IC 4686/IC 4687. metric extinctions over the Pa emitting regions for our The H images of IC 4734, and NGC 7130 (Dopita et sample of galaxies are given in the second column of Ta- al. 2002), as well as of IC 5179 (see Lehnert & Heckman ble 6. 1995) and NGC 6701 (see Marquez, Moles, & Masegosa A comparison with the spectroscopic extinctions in Ta- 1996) show H ii region and diuse H emission over scales bles 4 and 5 shows that the photometric extinctions are larger than the FOV of the Pa images, and thus our comparable to or slightly smaller than the extinctions to extinction-corrected Pa luminosities are lower limits the gas for the galaxies in common. However, in most to the total emission. If the average extinction of H ii cases, the photometric extinctions are derived for larger regions decreases for increasing galactocentric distances areas than the regions covered by the spectroscopic ob- (Calzetti et al. 2005), as in spiral galaxies, the contri- servations. Since there is evidence in spiral galaxies that butions from the large-scale component not covered by the extinction tends to decrease with distance from the the NICMOS images derived from the H images will be galaxy center (Calzetti et al. 2005), the average photo- upper limits. Alonso-Herrero et al. 17

We have retrieved the available MIPS 24 m images found for the inner region of M51 suggesting that dier- from the Spitzer archive and simulated the NICMOS ences among dierent types of galaxies are likely to be FOV to estimate the contribution from the large-scale present. emission. At 24 m the NICMOS FOV ( 2000 2000) Fig. 3 shows a comparison between the IR and covers most of the emission (> 85%) for the majority the monochromatic IRAS 12 m luminosities and the of the galaxies of the sample, except for those listed in extinction-corrected Pa luminosities for our sample of Table 7. For those galaxies with available H imaging, LIRGs. The 12m luminosities for close interacting sys- the agreement on the fraction of extended emission de- tems are based on Surace et al. (2004). When they could rived from the 24 m imaging is reasonably good. We not obtain IRAS uxes for the individual components of conclude that, with a few exceptions, the Pa detected close interacting systems, the 12 m and IR luminosities by NICMOS is representative of the totals for our sample of each component were assumed to have a ratio similar of galaxies. to that of their Pa luminosities. 9. In addition to our sample of LIRGs, we have compiled STAR FORMATION RATES OF LIRGS a small comparison sample of normal galaxies with avail- Both the IR emission and NLy are good tracers of the able NICMOS Pa imaging observations (not corrected massive SFR in galaxies. NLy is usually derived from hy- for extinction) from Boker et al. (1999). The normal drogen recombination lines, and has the advantage that galaxies were chosen so that the NICMOS Pa obser- it traces the current massive SFR, almost independently vations covered the majority of the ionized hydrogen of the previous star formation history of the galaxy (Ken- emission. If the values of the extinction inferred from nicutt 1998). The main disadvantage in deriving SFRs H/Pa line ratios for a few spiral galaxies and star- from NLy is determining AV and the fraction of escaping forming galaxies (AV 1 4 mag, see e.g., Maoz et al. photons from H ii regions. The former problem can be 2001; Quillen & Yukita 2001; Calzetti et al. 2005) are alleviated by using near-IR, mid-IR, and radio recombi- representative of our sample of normal galaxies, the ex- nation lines (e.g., H92, see Zhao et al. 1997; Roy et al. tinction correction for the Pa luminosity would only be 2005). The issue of the fraction of escaping photons is up to 0.2 dex. Finally to extend the luminosity range we still uncertain and is not considered here, but it has been have included in this comparison the four ULIRGs im- addressed both observationally (Beckman et al. 2000) aged in Pa by Murphy et al. (2001). For the ULIRGs and theoretically (Bland-Hawthorn & Maloney 1999). the Pa luminosities have been corrected for extinc- The IR luminosity of a starburst galaxy is due to UV tion using the AV values these authors derived from the emission (produced mainly by the young stellar popula- H/Pa line ratios. tion) absorbed by dust and re-emitted in the thermal IR. There is a tight correlation between both the IR and As discussed by Kennicutt (1998), the ecacy of the IR 12 m, and the extinction-corrected Pa luminosities, al- luminosity as a tracer of the SFR depends on the con- though some of the galaxies with deviating IR luminosi- tribution of the young stars to the heating of the dust, ties (marked in the gures), seem to follow better the and requires that all the UV light from massive stars be correlation when the 12 m luminosity is used. For ref- absorbed by the dust (i.e., AV > 1 mag, see also Calzetti erence we give in the last column of Table 6 the ratio et al. 2005). In the presence of high dust opacities, as in between the SFRs derived from the IR luminosity and LIRGs, the IR luminosity dominates the bolometric lu- the extinction-corrected Pa luminosity using the pre- minosity of the system and is the ultimate tracer of the scriptions of Kennicutt (1998). As can be seen from this SFR, but with a somewhat ill-determined time scale. table, for local LIRGs the SFRs derived from the number of ionizing photons are on average 0.20.3 dex lower than 9.1. Comparison between the mid-IR and IR those inferred from the total IR luminosity. This behav- luminosities and the extinction-corrected Pa ior is consistent with the tendency for the measured IR luminosities luminosity to include some contribution from older stars The mid-IR luminosities are routinely used to estimate (see below). the total IR luminosities and SFRs of galaxies at cosmo- 9.2. logical distances (e.g., Elbaz et al. 2002; Perez-Gonzalez The 24 m emission as a SFR tracer for dusty et al. 2005; Le Floc’h et al. 2005). Observationally star-forming galaxies the 12 m luminosity is found to be a good indicator We now examine in more detail the relation between of the total IR luminosity of local galaxies (e.g., Rush, mid-IR 24 m and Pa luminosities. The monochro- Malkan, & Spinoglio 1993; Elbaz et al. 2002; Takeuchi matic IRAS 25 m luminosities for our sample of LIRGs, et al. 2005). The advantage in using mid-IR luminosi- normal galaxies, and ULIRGs have been converted to ties as indicators of the SFR is that they are not aected monochromatic 24 m luminosities using the Dale & by the contribution from cold dust heated by old stars Helou (2002) models with appropriate indices for the that may dominate the far-IR luminosities. Spitzer ob- dust distribution (in their notation we use = 1.5 and servations of nearby normal galaxies are now showing = 2 for LIRGs and normal galaxies, respectively). In that there is a good correlation between the Pa or H addition we use Spitzer/MIPS 24 m observations of re- luminosity (corrected for extinction) and the 24 m lu- solved star-forming regions within the central 6 kpc re- minosity of H ii knots and H ii regions (e.g., M51 Calzetti gion of M51 from Calzetti et al. (2005), and for M81 et al. 2005; M81 Perez-Gonzalez et al. 2006), indicating (entire galaxy) from Perez-Gonzalez et al. (2006). For that the latter luminosity could also be a good poten- M51, Calzetti et al. (2005) have derived the extinction tial SFR tracer. Calzetti et al. (2005) noted, however, corrections from H/Pa line ratios, and for M81 we that the 24 m luminosity to SFR ratios of UV-selected transform the extinction-corrected (based on AV derived starbursts and ULIRGs deviate from the average value from the Balmer decrement or radio) H luminosities 18 Local LIRGs

Fig. 3.— Left panel: Comparison between the IR and the extinction-corrected Pa luminosities. For NGC 7469 we have derived the extinction-corrected Pa luminosities of the star-forming ring from the Br imaging data of Genzel et al. (1995). Note that three ULIRGs of the Murphy et al. (2001) sample are not detected at 12 m, and thus their LIR are shown as upper limits. Right panel: Comparison between the monochromatic 12 m and the extinction corrected Pa luminosities.

(see Perez-Gonzalez et al. 2006 for more details) to Pa t including normal galaxies, LIRGs, and ULIRGs. The assuming Case B recombination. The inclusion of data t to the M81 H ii regions alone provides a similar slope, for resolved star-forming regions within M51 and M81 essentially equal to unity (0.987 0.064). allows us to extend the range of the mid-IR-Pa relation As also noted by Calzetti et al. (2005) there are varia- down three orders of magnitude. tions in the mid-IR to Pa ratios from galaxy to galaxy, As can be seen from Fig. 4 (left panel), the LIRGs, clearly shown in Fig. 4 (right panel). In particular, the ULIRGs, and normal galaxies seem to continue the re- L(24 m)/L(Pa)corr ratios of the M81 H ii regions ap- lation between the extinction-corrected Pa luminosity pear to follow a dierent trend in terms of the Pa lumi- and the 24 m luminosity observed for the M51 central nosity than the rest of galaxies and the M51 H ii regions. H ii knots. We also show in this gure data for galax- (Perez-Gonzalez et al. 2006 attribute the large scatter ies from the Universidad Complutense de Madrid sam- of L(24 m)/L(Pa)corr in M81 to uncertain extinction ple (UCM, Perez-Gonzalez et al. 2003) and the Nearby corrections). The variation of the L(24 m)/L(Pa)corr Field Galaxy Sample (NFGS, Kewley et al. 2002) with ratio with the Pa luminosity (i.e., SFR) between the available H imaging, and extinction corrections derived M51 H ii knots and our LIRGs and ULIRGs is simi- from the Balmer decrement. These galaxies also follow lar to the L24 m/SFR ratios found by Calzetti et al. the trend but with a larger scatter, as their measure- (2005). In fact, the gap between the M51 H ii knots ments are from H imaging rather than Pa. The M81 and the LIRGs/ULIRGs seen in our gure is occupied H ii regions, on the other hand, follow a similar linear re- by their UV-selected sample of starbursts, and the UCM lation, but the relation appears to be oset with respect and NFGS galaxies. We note however that we nd a to that of M51, normal galaxies, and LIRGs (see discus- more pronounced variation of the L(24 m)/L(Pa)corr sion by Perez-Gonzalez et al. 2006). This behavior could ratio because we use the Pa luminosity as a proxy for arise from a lower UV absorption eciency in the rela- the SFR for LIRGs, whereas they use the IR luminosity tively low luminosity and lightly obscured H ii regions in (see previous section, and Table 6). M81. The best t to the extinction-corrected Pa vs. 24 m 9.2.1. Deeply embedded sources luminosity relation for resolved star-forming regions of M51, normal galaxies (with available Pa imaging), The deviation from strict proportionality in the LIRGs (excluding IC 860), and ULIRGs is: L(24 m) vs. L(Pa)corr relation in dusty systems probably results from the eects of extinction. LIRGs and ULIRGs are known to contain highly obscured regions, log L(24m) = (3.5530.516)+(1.1480.013)log L(Pa)corrusually associated with the nuclei of the galaxies, that are (1) optically thick in the optical, and even in the near- and where the luminosities are in erg s1, and the 24 m lu- mid-IR (e.g, Genzel et al. 1995; 1998; Alonso-Herrero minosity is computed as the monochromatic value, i.e., et al. 2000; Doyon et al. 1994; Kotilainen et al. 1996; from f . For comparison the t to the M51 H ii regions Zhou et al. 1993). There is also indirect evidence based alone gives a slightly smaller slope (1.088 0.061, see on the observed ratios of mid-IR ne-structure emission also Calzetti et al. 2005), but within the errors of the lines that many of the youngest stars in massive star- Alonso-Herrero et al. 19

Fig. 4.— Left panel. Monochromatic 24 m luminosities vs. extinction-corrected Pa luminosities for our sample of LIRGs, as well as normal galaxies and ULIRGs (those shown in Fig. 3). We also show in this comparison data for individual star-forming regions within two nearby galaxies with Spitzer/MIPS 24 m observations. For M51 the Calzetti et al. (2005) Pa observations cover the central 6 kpc of the galaxy. For M81 (see Perez-Gonzalez et al. 2006) the H observations include the entire galaxy, and have been converted to Pa luminosities assuming Case B recombination. The solid line is the derived t to the M51 knots, normal galaxies, LIRGs (excluding IC 860), and ULIRGs (see text). For comparison the dashed line shows the t to the M51 H ii knots given in Calzetti et al. (2005). Note how star-forming galaxies from the UCM and NFGS surveys (grey dots, not included in the t) also follow the trend. Right panel. Ratio of the 24 m to extinction-corrected Pa luminosities vs. the extinction-corrected Pa luminosity. We only show galaxies with Pa imaging data, that is, the UCM and NFGS galaxies are excluded. Symbols are as in left panel. bursts may still be embedded in ultra-compact high den- larger than it is. sity H ii regions (Rigby & Rieke 2004) and thus hidden The high extinction indicated for the LIRGs, the pos- from us by large amounts of extinction. sibility that the extinction is actually higher than our The departures from a constant L(24 m)/L(Pa)corr estimates, and the likelihood that some of the youngest ratio may be linked to dierences in the physical con- stars are still embedded in ultracompact H ii regions are ditions in regions experiencing intense star formation. all consistent with our nding of a slope to the L(24m) In particular, the average extinction for the H ii regions vs. L(Pa)corr relation signicantly larger than unity. in M81 is AV 0.8 mag (Perez-Gonzalez et al. 2006), We believe that the physical explanation for this behav- whereas for the H ii knots of M51 is AV 3.5 mag ior is that the dust competes increasingly eectively for (Calzetti et al. 2005), and for our sample of LIRGs the ionizing photons and ultraviolet continuum photons in average extinctions over the Pa emitting regions are very heavily obscured systems, so that there is a trend AV 2 6 mag. for an increasing fraction of the luminosity to emerge in Since the estimated extinctions for the M51 H ii re- the IR. A surprising result, however, is that this trend gions and LIRGs are not too dierent, it is possible that is so similar from galaxy to galaxy. For reasonably dust we are underestimating the extinction in our sample of embedded galaxies and H ii regions, there is a linear em- LIRGs. As shown for Arp 299 by Alonso-Herrero et al. pirical relationship between IR and hydrogen recombina- (2000) and NGC 7469 by Genzel et al. (1995), the J H tion luminosity with very small scatter that holds over continuum colors tend to underestimate the extinction more than four decades in luminosity. in very dusty systems. Thus, it is possible that the pho- 9.2.2. tometric extinctions derived for some of the galaxies in Presence of AGN emission our sample are only lower limits to the true extinction The presence of a bright AGN could make a signi- (e.g., compare the photometric and spectroscopic extinc- cant contribution to the mid-IR luminosity of the sys- tions to the central regions of IRAS 171381017 and tem. Approximately 25% of the galaxies are classied NGC 2388). Particularly, this may be the case for the as AGN (Seyfert or LINER) on the basis of optical line Pa emission in highly inclined dusty galaxies (those de- ratios (see Table 1). Three additional LIRGs (ESO 320- scribed in §5.5). We note also that not all of the most lu- G030, IC 860, and Zw 049.057) present megamaser emis- minous IR systems in our sample are dominated by such sion (Norris et al. 1986; Baan, Haschick, & Uglesich extincted sources, for instance, the star formation prop- 1993). In extragalactic sources megamaser emission is erties of the merger NGC 1614 can be explained with generally associated with a strong compact radio con- AK ' 0.5 mag (see Alonso-Herrero et al. 2001). For- tinuum source, and about 70% of megamaser sources are tunately, the relative insensitivity of Pa to extinction classied as AGN or composite sources from optical spec- makes these dierences relatively unimportant in esti- troscopy (Baan et al. 1998), or tend to appear in galaxies mating the SFR, or the scatter in Fig. 4 would be much with warmer IRAS colors (Baan et al. 1993). Includ- 20 Local LIRGs ing the galaxies with megamaser emission the fraction of there are minimal issues with very heavily obscured re- AGN in our sample is consistent with that derived by gions that contribute no H). Assuming Case B recom- other works (e.g., Veilleux et al. 1995; Tran et al. 2001) bination, the relation between SFR and H quoted by 1 for the same IR luminosity range. Kennicutt (1998) can be converted to SFR(M yr ) = For NGC 7469 Genzel et al. (1995) estimated that 6.79 1041 L(Pa) (erg s1). From these considera- the nucleus contributes up to 40% of IRAS IR luminos- tions, we derive the following relation between the SFR ity (see also Soifer et al. 2003). For the B1 nucleus rate and mid-IR luminosity for luminous, dusty galaxies (that classied as a Seyfert) of NGC 3690 Keto et al. is: (1997) estimated a contribution of 20 30% in the mid- IR. For the rest of the galaxies in our sample containing 1 38 1 0.871 an AGN, only in the case of IC 4518W we nd that the SFR(M yr ) = 8.45 10 (L(24 m)/(erg s )) . Seyfert nucleus dominates the observed mid-IR emission (3) (see Alonso-Herrero et al. 2006 in preparation). This relation is analogous to the widely used relation Kewley et al. (2000) have argued, based on the radio between SFR and IR luminosity (Kennicutt 1998) but it properties of a sample of warm IR galaxies with IR lumi- is not aected by the uncertain contribution to the total nosities similar to our sample, that in the cases where an IR luminosity of a galaxy of dust heating from old stars. AGN is detected spectroscopically, it is rarely the dom- 10. SUMMARY inant power source for the IR luminosity. In most cases in the galaxies in our sample containing an AGN there is We have analyzed HST/NICMOS 1.1 1.89 m con- extended H ii emission over large scales (e.g., NGC 7130; tinuum and Pa emission line observations of a volume NGC 3690), so it is likely that the AGN contribution limited sample of 30 local universe LIRGs (log LIR = to the total IR luminosity of the system is signicantly 11 11.9 [L]). The galaxies have been selected so that smaller than in the case of NGC 7469 and IC 4518W. the Pa emission line could be observed with the NIC- MOS F190N narrow-band lter (2800 < v < 5200 km 1 9.2.3. The empirical calibration s ). The sample comprises approximately 80% of all the LIRGs in the RBGS in the above velocity interval, We can use the relation in equation (1) for empirical and is representative of local LIRGs in general. The NIC- estimates of the SFR in dusty environments. The most MOS observations cover the central ' 3.37.2 kpc of the direct application is to calculate the intrinsic extinction- galaxies. We nd the following: corrected Pa luminosity from the monochromatic luminosity at 24 m. Inverting equation (1), we nd: 1. The most common morphological continuum fea-

0.871 tures in the central regions are bright star clusters, L(Pa)corr = 1244 L(24m) . (2) and large-scale spiral arms, sometimes extending Kennicutt (1998) derived a relationship between H down to the inner kpc (mini-spirals). In highly luminosity and the SFR. He also converted this expres- inclined systems there are dust lanes crossing the sion to one between total IR luminosity and SFR in dusty disks of the galaxies and in some cases, even hiding galaxies, and this latter expression has been used widely the nucleus of the galaxy. The “extreme” (per- to interpret IR observations. We will now update the turbed) morphologies commonly seen in ULIRGs expression based on the results reported in this paper. are only observed in some of the most IR luminous First, we use monochromatic 24 m luminosities in place examples in our sample. of total IR luminosities. Kennicutt (1998) assumed that 2. Approximately half of the LIRGs show compact the great majority of the luminosity from young stars ( 1 2 kpc) Pa morphologies in the form of would be absorbed by dust and reradiated in the far- nuclear star formation rings, mini-spiral structure, IR in deriving his relationship. Although this assump- and emission associated with the nucleus of the tion is likely to be correct, observationally there may be galaxy. The typical observed (not corrected for other contributions to the total IR luminosity from older extinction) H surface brightnesses are between populations of stars or other luminosity sources (e.g., re- 2 10 1041 erg s1 kpc2, although the most ex- view by Tus & Popescu 2005). Therefore, a true total treme cases, for instance the ring of star formation IR luminosity measurement may overestimate the recent of NGC 1614, can reach ' 60 1041 erg s1 kpc2. star formation in a galaxy, although estimates based on IRAS measurements alone are less subject to these is- 3. The remaining half of the galaxies show Pa emis- sues because they only poorly sample the output of the sion extending over scales of ' 3.3 7.2 kpc and cold dust (because the longest band is at 100m). Sec- larger, with bright H ii regions in the spiral arms ond, Kennicutt assumed a direct proportionality between (face-on systems) or along the disks of the galaxies the H luminosity and the total IR. We nd empirically (edge-on systems), with or without bright nuclear that the increasing absorption eciency in increasingly emission. The H surface brightnesses are about luminous and obscured galaxies leads to a deviation from one order of magnitude fainter than in galaxies with strict proportionality toward increasing IR output with compact Pa emission. increasing star-forming luminosity. We begin by calibrating the mid-IR/Pa relation using 4. Most of the LIRGs show a population of numerous the H ii regions in M51. The extinction to these regions is bright H ii regions, and in about half of them the roughly 2 magnitudes at H. Since the morphologies are typical H ii region (the median H luminosity) is similar between H and Pa (Quillen & Yukita 2001), a as bright as or brighter than the giant H ii region straightforward extinction correction is appropriate (e.g., 30 Dor. The most IR (as well as in H) luminous Alonso-Herrero et al. 21

galaxies tend to host the brightest median and rst- s1))0.871 ranked H ii regions. 5. The extinctions to the gas (derived from the H/Pa and Pa/Br line ratios) and to the stars (derived from the mF110W mF160W colors) are on Based on observations with the NASA/ESA Hubble average AV 2 6 mag over the Pa emitting Space Telescope at the Space Telescope Science Institute, regions covered by the HST/NICMOS images. which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 6. There exists a good correlation between the mid-IR 5-26555. This research has made use of the NASA/IPAC 24 m and the extinction-corrected Pa luminosi- Extragalactic Database (NED), which is operated by the ties of LIRGs, ULIRGs, normal galaxies, and H ii Jet Propulsion Laboratory, California Institute of Tech- regions of M51, covering nearly ve decades in lu- nology, under contract with the National Aeronautics minosity. This suggests that the mid-IR luminosity and Space Administration (NASA). of galaxies undergoing dusty, intense star forma- We thank an anonymous referee for a careful reading tion is a good indicator of the SFR. The increasing of the manuscript and useful suggestions. We would like L(24 m)/L(Pa)corr ratio for more luminous Pa to thank D. Calzetti and R. Kennicutt for enlightening emitters (that is, galaxies with higher SFRs) may discussions. We are grateful to T. Hattori for providing be due to the presence of deeply embedded sources us with H images of the galaxies. We thank T. Daz- in the youngest star-forming regions for which we Santos for helping us with the Spitzer/MIPS images of have underestimated the extinction, even at near- the sample of LIRGs. AAH and LC acknowledge sup- IR wavelengths. port from the Spanish Programa Nacional de Astronoma y Astrofsica under grant AYA2002-01055 and Plan Na- 7. Analogous to the widely used relation between SFR cional del Espacio ESP2005-01480, and PGPG from the and total IR luminosity (Kennicutt 1998), we show Spanish Programa Nacional de Astronoma y Astrofsica that SFRs can be determined accurately in lumi- under grant AYA 2004-01676. This work has been funded nous, dusty galaxies as: by NASA grant HST-GO-10169 and by NASA through 1 38 SFR(M yr ) = 8.45 10 (L(24m)/(erg contract 1255094 issued by JPL/Caltech.

REFERENCES Alonso-Herrero, A., Rieke, G. H., Rieke, M. J., & Scoville N. Z. Genzel, R., Weitzel, L., Tacconi-Garman, L. E., Blietz, M., 2000, ApJ, 532, 845 Cameron, M., Krabbe, A., Lutz, D., & Sternberg, A. 1995, ApJ, Alonso-Herrero, A., Engelbracht, C. W., Rieke, M. J., Rieke, G. 444, 129 H., & Quillen, A. C. 2001, ApJ, 546, 952 Genzel, R. et al. 1998, ApJ, 498, 579 Alonso-Herrero, A., & Knapen, J. H. 2001, AJ, 122, 1350 Genzel, R. et al. 2001, ApJ, 563, 527 Alonso-Herrero, A., Rieke, G. H., Rieke, M. J., & Scoville, N. Z. Gehrz, R. D., Sramek, R. A., & Weedman, D. W. 1983, ApJ, 267, 2002, AJ, 124, 166 551 Arribas, S., Bushouse, H., Lucas, R. A., Colina, L, & Borne, K. D. Goldader, J. D., Joseph, R. D., Doyon, R., & Sanders, D. B. 1997a, et al. 2004, AJ, 127, 2522 ApJS, 108, 449 Baan, W. A., Haschick, A. D., & Uglesich, R. 1993, ApJ, 415, 140 Goldader, J. D., Joseph, R. D., Doyon, R., & Sanders, D. B. 1997b, Baan, W. A., Salzer, J. J., & LeWinter, R. D. 1998, ApJ, 509, 633 ApJ, 474, 104 Beckman, J. E., Rozas, M., Zurita, A., Watson, R. A., & Knapen, Hattori, T., Yoshida, M., Ohtani, H., Sugai, H., Ishigaki, T., Sasaki, J. H. 2000, AJ, 119, 2728 M., Hayashi, T., Ozaki, S., Ishii, M., & Kawai, A. 2004, AJ, 127, Bell, E. F. et al. 2005, ApJ, 625, 23 736 Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393 Hummer, D. G., & Storey, P. J. 1987, MNRAS, 224, 801 Bland-Hawthorn, J., & Maloney, P. R. 1999, ApJ, 510, L33 Kennicutt, R. C. Jr., Edgar, B. K., & Hodge, P. W. 1989, ApJ, Boker, T. et al. 1999, ApJS, 124, 95 337, 761 Boker, T., Sarzi, M., McLaughlin, D. E., van der Marel, R., Rix, Kennicutt, R. C. Jr. 1998, ARA&A, 36, 189 H.-W., Ho, L. C., & Shields, J. C. 2004, AJ, 127, 105 Keto, E. et al. 1997, ApJ, 485, 598 Bushouse, H. A., Borne, K. D., Colina, L., Lucas, R. A., Rowan- Kewley, L. J., Heisler, C. A., Dopita, M. A., Sutherland, R., Norris, Robinson, M., Baker, A. C., Clements, D. L., Lawrence, A., & R. P., Reynolds, J., & Lumsden, S. 2000, ApJ, 530, 704 Oliver, S. 2002, ApJS, 138, 1 Kewley, L. J., Heisler, C. A., Dopita, M. A., & Lumsden, S. 2001, Calzetti, D., Kinney, A, L., & Storchi-Bergmann, T. 1994, ApJ, ApJS, 132, 37 429, 582 Kewley, L. J., Geller, M. J., Jansen, R. A., & Dopita, M. A. 2002, Calzetti, D. et al. 2005, ApJ, 633, 871 AJ, 124, 3135 Clements, D. L., Sutherland, W. J., McMahon, R. G., & Saunders, Kim, D.-C., Sanders, D.B., Veilleux, S., Mazzarella, J.M., & Soifer, W. 1996, MNRAS, 279, 477 B. T. 1995, ApJS, 98, 129 Colina, L., Arribas, S., & Clements, D. 2004, ApJ, 602, 181 Kotilainen, J. K., Moorwood, A. F. M., Ward, M. J., & Forbes, D. Colina, L., Borne, K., Bushouse, H., Lucas, R. A., Rowan- A. 1996, A&A, 305, 107 Robinson, M., Lawrence, A., Clements, D., Baker, A., & Oliver, Lagache, G., Puget, J.-L., & Dole, H. 2005, ARA&A, 43, 727 S. 2001, ApJ, 563, 546 Le Floc’h, E. et al. 2005, ApJ, 632, 169 Corbett, E. A., Kewley, L., Appleton, P. N., Charmandaris, V., Lehnert, M. D. & Heckman, T. M. 1995, ApJS, 97, 89 Dopita, M. A., Heisler, C. A., Norris, R. P., Zezas, A., & Marston, Leitherer, C. et al. 1999, ApJS, 123, 3 A. 2003, ApJ, 583, 670 Lpari, S., Daz, R., Taniguchi, Y., Terlevich, R., Dottori, H., & Dale, D. A., & Helou, G. 2002, ApJ, 576, 159 Carranza, G. 2000, AJ, 120, 645 Dopita, M. A., Pereira, M., Kewley, L. J., & Capaccioli, M. 2002, Lpari, S., Daz, R. J., Forte, J. C., Terlevich, R., Taniguchi, Y., ApJS, 143, 47 Aguero, M. P., Alonso-Herrero, A., Mediavilla, E., & Zepf, S. Doyon, R., Joseph, R. D., & Wright, G. S. 1994, ApJ, 421, 101 2004, MNRAS, 354, L1 Elbaz, D., Cesarsky, C. J., Chanial, P., Aussel, H., Franceschini, McLeod, K. K., Rieke, G. H., Rieke, M. J., & Kelly, D. M. 1993, A., Fadda, D., & Chary, R. R. 2002, A&A, 384, 848 ApJ, 412, 111 Garca-Marn, M., Colina, L, Arribas, S., Alonso-Herrero, A., & Maoz, D., Barth, A. J., Ho, L. C., Sternberg, A., & Filippenko, A. Mediavilla, E. 2006, ApJ, submitted V. 2001, AJ, 121, 3048 22 Local LIRGs

Marquez, I., Moles, M., & Masegosa, J. 1996, A&A, 310, 401 Scoville, N. Z. et al. 2000, AJ, 119, 991 Martini, P., Regan, M. W., Mulchaey, J. S., & Pogge, B. W. 2003, Scoville, N. Z., Poletta, M., Ewald, S., Stolovy, S. R., Thompson, ApJ, 589, 774 R., & Rieke, M. J. 2001, AJ, 122, 3017 Melbourne, J. & Salzer, J. J. 2002, AJ, 123, 2302 Sekiguchi, K. & Wolstencroft, R. D. 1992, A&A, 255, 581 Melbourne, J., Koo, D. C., & Le Floc’h, E. 2005, ApJ, 632, 65 Shi, Y., Rieke, G. H., Papovich, C., Perez-Gonzalez, P. G., & Le Miles, J. W., Houck, J. R., Hayward, T. L., & Ashby, M. L. N. Floc’h, E. 2006, ApJ, in press (astro-ph/0603453) 1996, ApJ, 465, 191 Soifer, B. T., Sanders, D. B., Madore, B. F., Neugebauer, G., Murphy, T. W., Soifer, B. T., Matthews, K., & Armus, L. 2001, Danielson, G. E., Elias, J. H., Lonsdale, C. J., & Rice, W. L.s ApJ, 559, 201 1987, ApJ, 320, 238 Murphy, T. W. Jr., Armus, L., Matthews, K., Soifer, B. T., Soifer, B. T., Neugebauer, G. Matthews, K., Egami, E., Mazzarella, J. M., Shupe, D. L., Strauss, M. A. & Neugebauer, Weinberger, A. J., Ressler, M., Scoville, N. Z., Stolovy, S. R., G. 1996, AJ, 111, 1025 Condon, J. J., & Becklin, E. E. 2001, AJ, 122, 1213 Norris, R. P., Whiteoak, J. B., Gadner, F. F., Allen, D. A., & Soifer, B. T., Bock, J. J., Marsh, K., Neugebauer, G., Matthews, Roche, P. F. 1986, MNRAS, 221, 51P K., Egami, E., & Armus, L. 2003, AJ, 123, 146 Papovich, C., Egami, E., Le Floc’h, E., Perez-Gonzalez, P., Rieke, Sugai, H., Davies, R. I., Malkan, M. A., McLean, I. S., Usuda, T., G., Rigby, J., Dole, H., & Rieke, M. 2005, in press, proceedings & Ward, M. J. 1999, ApJ, 527, 778 of the STScI May 2004 Symposium, ”Planets to Cosmology: Surace, J. A., Sanders, D. B., & Mazzarella, J. M. 2004, AJ, 127, Essential Science in Hubble’s Final Years” (astro-ph/0408454) 3235 Perez-Gonzalez, P. G., Zamorano, J., Gallego, J., Aragon- Surace, J. A., Sanders, D. B., & Evans, A. S. 2000, ApJ, 529, 170 Salamanca, A., & Gil de Paz, A. 2003, ApJ, 591, 827 Takeuchi, T. T., Buat, V., Iglesias-Paramo, J., Boselli, A., & Perez-Gonzalez, P. G. et al. 2005, ApJ, 630, 82 Burgarella, D. 2005, A&A, 432, 423 Perez-Gonzalez, P. G. et al. 2006, submitted Tran, Q. D. et al. 2001, ApJ, 552, 527 Quillen, A. C., McDonald, C., Alonso-Herrero, A., Lee, A., Shaked, Tus, R. J., & Popescu, C. c. 2005, AIP Conf. Ser. 216, 84 S., Rieke, M. J., & Rieke, G. H. 2001, ApJ, 547, 129 van den Broek, A. C. et al. 1991, A&AS, 91, 61 Quillen, A. C., & Yukita, M. 2001, AJ, 121, 2095 Veilleux, S., Kim, D.-C., Sanders, D. B., Mazzarella, J. M., & Rieke, G. H. & Low, F. J. 1972, ApJ, 176, L95 Soifer, B. T. 1995, ApJS, 98, 171 Rieke, G. H. & Lebofsky, M. J. 1985, ApJ, 288, 618 Veron-Cetty, M.-P. & Veron, P. 2001, A&A, 374, 92 Rieke, G. H., Loken, K., Rieke, M. J., & Tamblyn, P. 1993, ApJ, Witt, A. N. & Gordon, K. D. 2000, ApJ, 528, 799 412, 99 Wu, H., Zou, Z. L., Xia, X. Y., & Deng, Z. G. 1998, A&AS, 132, Rigby, J. R., & Rieke, G. H. 2004, ApJ, 606, 237 181 Roy, A. L., Goss, W. M., Mohan, N. R., & Anantharamaiah, K. R. Wynn-Wiliams, C. G., & Becklin, E. E. 1993, ApJ, 412, 535 2005, A&A, 435, 831 Zhao, J.-H., Anantharamaiah, K. R., Goss, W. M., & Viallefond, Rush, B., Malkan, M. A., & Spinoglio, L. 1993, ApJS, 89, 1 F. 1997, ApJ, 482, 186 Sanders, D. B. & Mirabel, I. F. 1996, ARA&A, 34, 749 Zheng, X. Z., Hammer, F., Flores, H., Assemat, F., & Pelat, D. Sanders, D. B., Soifer, B. T., Elias, J. H., Neugebauer, B., & 2004, A&A, 421, 847 Matthews, K. 1988, ApJ, 328, L35 Zhou, S., Wynn-Williams, C. G. & Sanders, D. B. 1993, ApJ, 409, Sanders, D. B., Egami, E., Lpari, S., Mirabel, I. F., & Soifer, B. 149 T. 1995, AJ, 110, 1993 Sanders, D. B., Mazzarella, J. M., Kim, D.-C., Surace, J. A., & Soifer, B. T. 2003, AJ, 126, 1607 Satyapal, S., Watson, D. M., Pipher, J. L., Forrest, W. J., Fischer, J., Greenhouse, M. A., Smith, H. A., & Woodward, C. E. 1999, ApJ, 516, 704